|
from PIL import Image |
|
import torch |
|
import torch.nn.functional as F |
|
import numpy as np |
|
from dkm.utils.utils import tensor_to_pil |
|
import cv2 |
|
from dkm import DKMv3_outdoor |
|
|
|
device = torch.device("cuda" if torch.cuda.is_available() else "cpu") |
|
|
|
|
|
if __name__ == "__main__": |
|
from argparse import ArgumentParser |
|
|
|
parser = ArgumentParser() |
|
parser.add_argument("--im_A_path", default="assets/sacre_coeur_A.jpg", type=str) |
|
parser.add_argument("--im_B_path", default="assets/sacre_coeur_B.jpg", type=str) |
|
|
|
args, _ = parser.parse_known_args() |
|
im1_path = args.im_A_path |
|
im2_path = args.im_B_path |
|
|
|
|
|
dkm_model = DKMv3_outdoor(device=device) |
|
|
|
W_A, H_A = Image.open(im1_path).size |
|
W_B, H_B = Image.open(im2_path).size |
|
|
|
|
|
warp, certainty = dkm_model.match(im1_path, im2_path, device=device) |
|
|
|
matches, certainty = dkm_model.sample(warp, certainty) |
|
kpts1, kpts2 = dkm_model.to_pixel_coordinates(matches, H_A, W_A, H_B, W_B) |
|
F, mask = cv2.findFundamentalMat( |
|
kpts1.cpu().numpy(), |
|
kpts2.cpu().numpy(), |
|
ransacReprojThreshold=0.2, |
|
method=cv2.USAC_MAGSAC, |
|
confidence=0.999999, |
|
maxIters=10000, |
|
) |
|
|
|
|